01 Sep 2018
The True Value of a Tweet
Accounting for social media’s swayby Jen McFarland FlintTopics:
Image by John Ritter
Companies can spend millions curating their social media presence—with sometimes famously funny results, like the Twitter accounts for Wendy’s and Denny’s—but how relevant is that to revenues? Here, Associate Professor Leslie John and Professor Sunil Gupta suss out social media’s impact on the bottom line.
You and your colleagues have taken a critical look at the assumption that social media works by attracting followers, which exposes the brand and in turn leads to sales. What’s wrong with that line of thinking?
Leslie John: There’s some hyperbole out there about the value of “likes.” The evidence often cited to support that is to look at people who have liked Starbucks, for example, compared with people who have not. The assumption is that people who have liked Starbucks spend more money there. The critical flaw in that logic is that the people who liked the brand inherently like the brand in the offline sense of the word more than those who haven’t. And we don’t know what would have happened if they hadn’t liked the brand on Facebook.
We did a series of studies to try to disentangle that. And we found that if you just get more likes, those people don’t spend more than those who are not induced to like. That’s not to say that likes have no value. We think there are ways to potentially unlock their value, but it’s not this direct effect. We tend to get caught up on metrics that are easy to count, but there’s that saying: Not everything that can be counted counts, and not everything that counts can be counted.
Sunil Gupta: I think the excitement in social media happened because of the idea of earned media: You do something, it goes viral, now millions of people are talking about it, and all without you spending any money. Implicit in this assumption is the notion that you can make things go viral—and in fact, studies show that’s not easy. A 2012 study of millions of messages on Twitter and Yahoo! found that more than 90 percent of the messages didn’t diffuse at all, about 4 percent were shared only once, and less than 1 percent were shared more than seven times. It’s like winning a lottery.
So there are issues with the idea of virality, the self-selection involved in liking, and then there’s the contagion aspect that’s also part of the problem. There was an infamous study by Nicholas Christakis here at Harvard that said, basically, obesity is contagious: If my family and friends are obese, the likelihood of me being obese is much higher. Later studies refuted that claim because of the inherent problem of homophily, which simply means that birds of a feather flock together. If you buy a particular kind of music and we are Facebook friends, and I end up buying the same music later on, is it because you influenced me, or is it because we are friends with similar preferences? Going back to what Leslie said, we end up counting things that are easy to count, then we take a leap of faith without recognizing these fallacies in the conclusion.
What’s known about the value of Twitter in particular?
SG: I think a lot of this is vanity, to be honest. In some ways it’s like the old celebrity endorsements that we saw in television advertising. How many people buy Hanes underwear because Michael Jordan endorses it? Look at your own behavior. When was the last time you were influenced because somebody tweeted about a product? I think that research still needs to be done [as to] whether or not it can persuade somebody.
LJ: One way to use Twitter that may be undervalued is as a listening tool. There’s information in those rants. So instead of using it to persuade and project your message onto people, Twitter can be a tool to learn about your customers, as another way to get information about what they’re thinking.
SG: And also new product ideas: Are people talking about competitive products when they think about Nike shoes, for example? Are they talking about labor practices? Are they talking about cost? What are the issues on their minds? You can even use that minute-by-minute information to create a dynamic map. The caveat is that you might get a biased sample: In most cases, 5 percent or less of the total population will actually say something, and the rest is the silent majority.
So if you’re an executive, how do you pin a value to this tool?
LJ: There has to be a more careful study of that. But all of these digital tools are amazing for marketers and have enhanced our capability to test theories. You can much more easily do A/B testing and even more sophisticated tests to try to get at these questions—of online and offline behaviors—than you could in the past. It’s also just fun for us as experimentalists to try to come up with ingenious experimental designs.
SG: It’s an age-old story, but the challenge for all advertising is to connect the dots all the way to sales. Sales are a function of advertising but also price and the quality of a product, so you can’t blame or credit advertising for everything. And how do you ever calculate the cost-benefit analysis, especially if you’re a firm spending millions of dollars to build followers? That’s why we end up relying on these intermediate metrics, intuition, and anecdotal stories—and that’s where we go wrong.